面向时变机动目标的自适应航迹整体估计方法

Adaptive trajectory-oriented estimation estimation method for time-varying maneuvering targets

面对目标运动模型时变未知的复杂机动目标跟踪场景,采用确定性的状态空间模型会因模型失配而导致跟踪性能显著下降.为此,提出一种基于航迹整体估计的自适应跟踪方法.该方法将目标航迹整体建模为多项式时间序列,建立时间多项式模型表征其运动状态的演化过程,依据最小离差平方和准则对多项式系数进行矩阵形式的递归估计,从而在每个采样时刻实现对全时段航迹的整体估计,并进一步提出多项式系数的渐进解耦估计方法以降低运算复杂度.在此基础上,采用不同次数的多项式模型构建多通道的并行航迹整体估计器组,根据实时误差分析结果进行自适应通道选择,以应对时变机动目标复杂运动形式引起的单一模型失配问题.仿真结果表明,该方法在无法确知目标真实运动状态的情况下,能够实现自适应选择跟踪通道及模型参数调整,相较传统递推滤波方法,该整体估计方法具有更好的跟踪性能.

In the complex maneuvering target tracking scenario with unknown time-varying target motion model,the use of a fixed state-space model will significantly reduce the tracking performance due to model mismatch.In view of the above problem,this paper proposes an adaptive tracking method based on trajectory-oriented estimation estimation.In this method,the trajectory-oriented estimation of the target is modeled as a polynomial time series;the time polynomial model is established to characterize the evolution process of its motion state;the polynomial coefficients are recursively estimated in matrix form according to the minimum deviation sum of squares criterion,so as to achieve the global estimation of the full-time track at each sampling time,and furthermore,the asymptotic decoupling estimation method for polynomial coefficients is proposed to reduce the computational complexity.On this basis,the polynomial models of different degrees are used to construct a multi-channel parallel trajectory-oriented estimation estimator group,and adaptive channel selection is carried out according to the real-time error analysis results to deal with the single model mismatch problem caused by the complex motion pattern of time-varying maneuvering targets.Simulation results show that the proposed method can realize the selection of adaptive tracking channels and model parameter adjustment when the real motion state of the target cannot be ascertained,and that compared with the traditional recursive filtering method,this trajectory estimation method has better tracking performance.